I have been preaching the value and virtues of QbD (with PAT as the control mechanism, of course) for some time now. However, I still meet a large number of people who seem to have trouble seeing what QbD means in everyday terms. They don’t seem to comprehend the concept of modifying the process (in real time) to mirror current circumstances/conditions. So, let’s try another tack: one morning, while I was lying awake and waiting for the dogs to notice that fact, I was thinking about the “real meaning” of QbD. As I am the cook in our family (as was my dad), and addicted to the Food Network, I thought that QbD just might be explained in terms of cooking.
A top chef seldom measures exact amounts of flour, water, salt, pepper, and the like. He or she seasons to what he knows is the “correct” taste. When adding the ingredients, the chef knows that “one onion” or “two potatoes” are not fixed values in the real world. Since these are natural products, subject to soil, weather, and fertilizer levels, the actual sizes, flavors, and consistencies will vary. The professional chef will take the end product parameters into consideration and adjust the amounts put into the pot in order to achieve the correct flavor and texture of the plated dish.
Consider that, in almost every case, the ingredients we add to the API to generate a final dosage form are also natural products, and are subject to the same vagaries as the food we cook. Extrapolating from that fact, the question we need to ask is: If we routinely anticipate that food ingredients vary, why do we resist accepting that excipients vary? Why the fierce insistence on (blindly) following the (interpretation) of cGMPs that, once an SOP is written, cannot be changed, for any reason? Despite all the evidence to the contrary, too many companies still insist on fully characterizing only the API; they still apply only compendial methods (e.g., USP) for the excipients, which often comprise up to 95% or more of the final dosage form.
So, just as there is no such thing as a tomato or pepper USP-NF, there is no standard talc, or lactose, or starch. At least, there is no given consistency of physical properties. There may be many suppliers of excipients, each with its own source and packaging facilities, so there are often large variations in incoming raw materials. Unfortunately, the tests we routinely run only assure that the starch, lactose, or whatever contains acceptable levels of heavy metals, or residue(s) on ignition, or residual solvents. Any spectral test is run to show that, chemically, the material resembles a “standard” starch, or lactose, or talc. The physical tests, when run, are very broad and open ended: “no more than 1% is retained on a 100 mesh screen.” In short, there are no compendial tests for performance of a product.
This is more than interesting, since we insist on DQ/IQ/OQ/PQ for equipment employed in the lab or production. (The Design Qualification is becoming more important as we apply PAT and use customized equipment, but that is another column.) We are required to show, through the PQ portion, that a particular instrument (excipient?) is suitable for the application. It would be nice if the excipient manufacturers were bound by our rules, but they are not. Since they provide more of their products to other industries (lactose for food, cocoa butter for candies), we do not have the “hammer” to insist on performance parameters. Thus, we need to accommodate the reality of the situation and modify our needs (production parameters) to the quality of the excipients (tomatoes).
I find it interesting that, in light of decreasing profits and ending patents, so many companies are not embracing QbD concepts. Perhaps, if QbD were made as “easy as pie” they could see their way to embracing it more fully.
(Post-column comment: Just as 2000 was not the beginning of the 21st century, 2000 to 2009 was not the 00’s decade. Since we count from one through ten (at least, I don’t start counting with zero), 2001 was the first year of the decade that began the 21st century; the decade ends with 2010 and a new one starts with 2011. This sounds trivial, but shows that we don’t always pay attention to details and it shows in some assays and production problems.)